Download RB Buiatti

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Primary transcript wikipedia , lookup

Human genetic variation wikipedia , lookup

Minimal genome wikipedia , lookup

Nutriepigenomics wikipedia , lookup

Human genome wikipedia , lookup

Cre-Lox recombination wikipedia , lookup

Extrachromosomal DNA wikipedia , lookup

Synthetic biology wikipedia , lookup

Gene expression programming wikipedia , lookup

Genome (book) wikipedia , lookup

Koinophilia wikipedia , lookup

Genetic engineering wikipedia , lookup

Quantitative trait locus wikipedia , lookup

Mutation wikipedia , lookup

Vectors in gene therapy wikipedia , lookup

Epistasis wikipedia , lookup

Biology and consumer behaviour wikipedia , lookup

Genomics wikipedia , lookup

Deoxyribozyme wikipedia , lookup

Therapeutic gene modulation wikipedia , lookup

Site-specific recombinase technology wikipedia , lookup

Designer baby wikipedia , lookup

Gene wikipedia , lookup

Genome evolution wikipedia , lookup

Non-coding DNA wikipedia , lookup

Population genetics wikipedia , lookup

Genome editing wikipedia , lookup

Point mutation wikipedia , lookup

History of genetic engineering wikipedia , lookup

Helitron (biology) wikipedia , lookup

Artificial gene synthesis wikipedia , lookup

Microevolution wikipedia , lookup

Transcript
Chance vs. Necessity in Living Systems
29
Articles
Marcello Buiatti1 and Marco Buiatti 2
Chance vs. Necessity in Living Systems: A False Antinomy
1. Introduction
2. A Historical Review of the Order/Randomness Debate
3. The Contemporary Revolution in Biology and “The Benevolent Disorder”
4. Conclusions
Abstract. The concepts of order and randomness are crucial to understand
‘living systems’ structural and dynamical rules. In the history of biology, they
lay behind the everlasting debate on the relative roles of chance and determinism in evolution. Jacques Monod [1970] built a theory where chance
(randomness) and determinism (order) were considered as two complementary aspects of life. In the present paper, we will give an up to date version of
the problem going beyond the dichotomy between chance and determinism.
To this end, we will first see how the view on living systems has evolved from
the mechanistic one of the 19 th century to the one stemming from the most
recent literature, where they emerge as complex systems continuously evolving
through multiple interactions among their components and with the surrounding environment. We will then report on the ever increasing evidence
of “friendly” co-existence in living beings between a number of “variability
generators”, fixed by evolution, and the “spontaneous order” derived from
interactions between components. We will propose that the “disorder” generated is “benevolent” because it allows living systems to rapidly adapt to
changes in the environment by continuously changing, while keeping their
internal harmony.
Rivista di Biologia / Biology Forum 101 (2008), pp. 29-66.
30
Marcello Buiatti, Marco Buiatti
1. INTRODUCTION
All through the history of Biology one of the most relevant
sources of discussion has been the debate on the relative impact on
living systems of chance and necessity. In this paper we shall propose a new vision of this critical dilemma derived from an up to
date conception of the peculiar features of the “living state of
matter”. For this purpose we shall start by better defining the
meaning, in living systems, of the two apparently opposite concepts.
At variance with non living systems, living ones can be considered as “individuals”, as it is always possible to distinguish in them
the “internal” from the “external”.
All biological systems are composed of a number of elements,
often compartmentalized and connected to one another, isolated
from the exterior by some membrane, but continuously exchanging energy and matter with the environment. In the case of ecosystems, they are not limited by physical barriers but by the connections between different organisms. In other words all connected
components belong to one system, organisms coming from “outside” being defined as invasive individuals and species.
Components within individual living systems (molecules, cells,
organisms, populations, species, ecosystems) can assume a number
of configurations.
The level of “order” of a system can be measured through the
evaluation of the number of possible configurations and of their
relative probabilities. Within this frame, only one configuration is
allowed in wholly ordered system, maximum disorder being reached
when all possible configurations have the same probability to occur. The most direct way to measure the level of constraints to
randomness consists in estimating the ratio between maximum
entropy corresponding to equi-probable configurations (total entropy according to Atlan [1979]) and the entropy observed in the
systems under study (the observed entropy of Atlan). An observed
entropy significantly lower than the maximum entropy signals that
the different components of the system are correlated one with
each other, i.e. there is a significant amount of constraints to randomness.
Chance vs. Necessity in Living Systems
31
What is the role of order and randomness in the evolution of
living systems? What is the functional meaning of constraints to
randomness in the continuous fight for the permanence in the
“living state of matter” (as previously defined in Buiatti [2000];
Buiatti et al. [2004]; Buiatti and Buiatti [2004])? These questions
have always been at the core of biology, and since we, the humans,
are indeed living systems, they have always had an impact on our
general conception of life and of ourselves. This has provoked a
permanent and often misleading interaction between science and
ideology. On one hand biological data have been often used to
support ideological concepts and socio-political choices. On the
other, scientists have been themselves unconsciously influenced by
the changing “spirit of times” (“ZeitGeist”) acting as a filter or as
a pair of eyeglasses to be used when observing nature.
For a long time, the “spirit of times”, influenced by the industrial revolutions and a dominating deterministic ideology, has
pushed for the attribution of a positive “value” only to constraints
and “rules” or, in other terms to “seeds of order”, considering variability as disturbing noise. This has led to a rather unfortunate bias
in the discussion on the significance of randomness in life, often
reduced to an irrational fight over a supposed order/disorder antinomy.
In this paper, we shall try to overcome this historical antinomy
by showing that living systems are endowed both with ordered
features and a relevant amount of plasticity (disorder), and that
both components are fundamental for living systems to adapt to
the surrounding environment. To this aim, we will focus in the
first place on DNA, the hereditary material and therefore the “living memory” of biological dynamics.
We will first start with a brief overview of the history of the
discussion on the order/disorder antinomy since the discovery of
heredity laws by Mendel to our days. We will then propose a “solution” to this antinomy, based on the increasing experimental
evidence of “friendly” co-existence in living systems between a
number of “variability generators”, fixed by evolution, and the
“spontaneous order” derived from interactions between components.
32
Marcello Buiatti, Marco Buiatti
2. A HISTORICAL REVIEW
OF THE ORDER/RANDOMNESS DEBATE
1. From Mendel to the “Central Dogma of Molecular Genetics”
The relative contribution of order to the organization of living
systems has been for the aforementioned reason the object of a
long lasting debate particularly since the beginning of the industrial revolution and the explosion of “modern” science. The temporal connection between these two events should not be considered a mere coincidence. Human artifacts and particularly “machines” (the products of industry), are highly ordered systems, as
their components can only assume the configuration allowed by
the human project following which they have been constructed.
Moreover, machines are an ordered assemblage of independent
components in the sense that the removal of one of them does not
have any effect on the structure and function of the others. In
other words, a part of the machine as for instance a wheel in a car
is exactly the same when assembled with the other components as
it is before the assemblage. The situation is obviously different in
the case of living systems where all components are parts of networks where the extent of non additive interactions depends, as we
shall see later, on the number and the weight of links between any
given part and the others. Now, one of the basic methodological
innovations of “modern” science has been the simplification of a
complex system into its components to be studied one by one,
with the aim of obtaining the knowledge of the whole system
through the mere assemblage of the information thus gained. The
application of the method of “reduction” of the knowledge of a
whole to that of its components has been successfully applied to
non living systems since the 18th century in Physics and Chemistry
but was officially introduced in Biology only in the second half of
the 19th century when the “Manifesto of medical materialists” was
published (1848) by a group whose leaders were Von Helmholtz,
Du Bois Raymond, Brücke. The rational basis of the extension to
life of the use of the reductionist method has been, coherently
with the modern “spirit of time”, the principle of substantial equivalence of living and non living system. This principle implies that
Chance vs. Necessity in Living Systems
33
the laws of physics and chemistry are necessary and sufficient for
the total understanding of living systems which may therefore be
considered essentially as “living machines”. It is implicit in this
conception that Biology students should, whenever possible, dissect the systems to be investigated into components and analyze
them, eventually aiming, as physicists and chemists, at the development of mathematical hopefully “universal” laws. The mathematization of life becomes therefore the ultimate goal of Biology as it
had already happened since a long time in the case of other disciplines.
Now, in many cases mathematization of natural systems laws,
has been developed with the aim of predicting system dynamics,
an extremely useful exercise for humans, whose adaptation strategy
is largely based on the modification of nature possibly without
unwanted and unpredicted “collateral effects”. Given the self-reproductive capacity of living systems, long term prediction can
only be achieved through the discovery of mathematical laws of
heritability. Certainly this has not been the conscious aim of Gregor
Mendel when he started his experiments with peas. However, he
was a physicist, studied in Wien and had some loose but relevant
relationships with some of the founders of reductionism in Biology. With that cultural background (Orel [1983]), he decided to
analyze in peas the behavior of single characters discrete variants
(red or white color of the flower, rough or smooth seeds etc.), this
being the best way to obtain the numerical ratios of the variants
themselves appearing in subsequent generations of self-fertilization
of a cross between “pure lines”. These numbers may then open the
way to the mathematization in probabilistic terms of heritable
processes. The results obtained were perfectly adequate to Mendel’s expectations and seemed to confirm also in the field of heredity the high heuristic power of the reductionist method. Mendel
unfortunately never realized the real relevance of his data. The reason for his lack of understanding of the value of his own work was
again his training as a physicist inducing him to look for universal
mathematical laws, thus underestimating the complexity of his
object of study. With the aim of understanding whether the laws
developed for peas could be applied to other organisms and not
being a biologist, Mendel submitted his results to the critical re-
34
Marcello Buiatti, Marco Buiatti
view of a famous botanist of his period, von Naegeli. The answer
of the “referee” was moderately positive but he suggested repeating
the experiment on a plant different from the pea. Unfortunately
von Naegeli was not aware of the fact that the plant proposed for
the new experiments (Hyeracium) is not endowed with sexual reproduction. Therefore Mendel failed to obtain the same results
found in the previous experiments, considered them as essentially
irrelevant and unfortunately never got to know that he had started
one of the most important revolutions ever happened in Biology.
For this reason Mendel completely abandoned his studies of heredity and worked for the rest of his life on the behavior of bees
and on the mathematical laws of winds dynamics, as can be found
written on his grave. Mendel’s work, although known to a fair
number of scientists, was only reconsidered at the beginning of the
twentieth century, his “laws” being confirmed almost at the same
time by three biologists, a Dutch (Hugo De Vries), a German aristocrat living in the Austrian empire (Erich Tschermak von Seissenegg) and a Belgian (Carl Correns).
Let us now discuss with a little more detail the main concepts
developed by the Moravian scientist. According to Mendel’s results, heredity is controlled by discrete factors (“anlage”) present
each in two copies in every individual heritable complex. Due to
what De Vries later called “mutations”, two or more discrete variants (“alleles”) of each factor can be found in each population and
species. Alleles can be “dominant” or recessive and individuals
“homozygotes” (bearing two copies of only one allele) or “heterozygotes” (with one copy of two alleles). As every factor (“gene”)
was thought to carry, in a wholly deterministic way, the information for one character, the genotype (the assemblage of the alleles
and genes) of an organism would dictate without ambiguity the
corresponding phenotype.
Now, if we call “A” the dominant allele of one character (for
instance the red color of a flower) and “a” the recessive allele of
the same gene (for instance white), a heterozygous genotype would
be Aa, the two homozygous ones being AA and aa. For reproduction the organisms produce “gametes” (spermatozoa if males and
ovules if females in animals) containing now only one copy of
every gene and therefore only one allele. The ovules of heterozy-
Chance vs. Necessity in Living Systems
35
gote individuals may then be either A or a and the same would
hold true for the spermatozoa of a male. At fertilization one sperm
and one ovule would meet and give rise to the “zygote” from
which one member of the progeny would develop. Then, without
interferences by other factors, the probability of a sperm or an
ovule from our heterozygote containing A or a being the same
(50%), the probabilities of the three possible genotypes of progeny
members (AA, Aa, aa) would be respectively 1/4; 2/4; 1/4. Phenotypes, accordingly, given the dominance rule, will be then, in our
case 3/4 dominant (red) and 1/4 recessive (white). Summarizing
those results, the distribution of alleles to the progeny is for
Mendel completely random, the phenotypes of each individual of
the new generation being univocally determined by the randomly
acquired genotypes. To be more precise, every individual of a species has the same gene complement (the same “functions”) although in different variants (every pea has a flower color, white or
red), random assortment referring to the alleles, that is to gene
variants. In other words, the genetic complement of one individual
is for Mendel and his followers, as envisaged by Ernst Mayr, like a
“bean bag” (Mayr [1982]) from which alleles are randomly chosen
and consequently an organism is supposed to be like a machine
determined by one of the many equi-probable combinations of the
“bean” variants deriving from parental genotypes.
It is also noteworthy that the alleles whose distributions have
been analyzed by Mendel and the Mendelians had on purpose
been chosen to be discrete (red or white, rough or smooth etc.) for
the simple reason that discrete objects are easy to count in subsequent generations. Therefore all results and the derived theories
did not take into account the possible existence of genes controlling quantitative characters showing a continuous distribution of
values like for instance weight, height, length etc. This fact started
a very harsh debate in the early years of the twentieth century between the Mendelians and the school of a British statistician, the
Darwin relative Sir Francis Galton, an orthodox Darwinian and a
student of quantitative characters. The debate was particularly
centered on the relevance of Mendelian results for the understanding of evolution, the Galtonians strongly supporting the Darwinian view of evolution as a continuous process. The only compre-
36
Marcello Buiatti, Marco Buiatti
hensive theory opposed to the Galtonian view was proposed by
one of the “discoverers” of Mendel’s work at the beginning of the
twentieth century, Hugo De Vries, in his two-volume treatise, Die
Mutations Theorie, first published in 1902. In his masterpiece he
was the first to confirm in a large number of experiments Mendel’s
data and to extrapolate from them in a clear way the two concepts
of major relevance for evolutionary dynamics of the Mendelian
view, namely randomness and discreteness. He however, integrated
Mendel’s view with the new concept of “mutation” defined as a
discrete “jump” from one allele to another.
According to De Vries, mutations were totally random events
occurring all with the same frequency in time. Evolution therefore
would derive from the linear accumulation of mutations with time
without any interference of the environment. De Vries even wrote
a simple linear regression function correlating mutation numbers
with time, thus extending to evolution the mathematization
started by Mendel with his probabilistic treatment of genes and
characters behavior throughout generations. Evolutionary dynamics was then characterized as a cyclic occurrence of total “disorder”
(maximum entropy) due to the random assortment of alleles in
gamete production and fertilization followed by total order, the
development during life cycles. Development was envisaged as a
unique path of configurations changing from birth to death, completely determined “a priori” by the genotype. In synthesis, life
processes were thought to be controlled by a blind “bricoleur”
(Jacob [1970]) giving rise to random changes in otherwise deterministic machines.
This vision was in open contradiction with Darwin’s conception which attributed evolution to three interacting factors: environment controlled selection, use and disuse of organs and correlated variation. This last factor introduced the very interesting
concept of constraints to randomness deriving from the interaction between the parts of an organism and from the “rules” needed
for the maintenance of a sufficient level of dynamic harmony between components. According to this concept, clearly stated by
Darwin but not by neo-Darwinians, only mutations coherent with
the desired amount of harmony would be compatible with life and
therefore fixed by selection. The author of the “Origin”, wrote in
Chance vs. Necessity in Living Systems
37
1876 in “Variation in animals and plants under domestication”:
Although every variation is directly or indirectly caused by
changes in external conditions, we should never forget that the
nature of the organization on which we act is by far the most
important factor influencing the result.
The Mendelian-Galtonian controversy on the two sets of opposite concepts chance/necessity and continuous/discrete was solved
few decades after its inception, when a compromise was attempted
by the “Modern synthesis” as it was defined in the forties of the
twentieth century by Julian Huxley [1942]. In this theory, only
one of the Darwinian factors of evolutionary dynamics was re-introduced, namely natural selection. According to the neo-Darwinists, selection operates on single genes and alleles, still believed
to be as independent as Mayr’s beans in a bag, the “best” alleles
being chosen out of those present in a population. The choice is
based on the relative reproductive capacity conferred to individuals
by the different alleles. The reproductive capacity (average progeny
size of individuals carrying a given allele) is called “fitness” and
used as the only parameter for the evaluation of the level of adaptation. To take a classical example, in a population where the three
possible genotypes (AA, Aa, aa) are present, phenotypes are either
A or a due to the presence of dominance and, at least in the classical version of the theory where interaction between alleles was
not taken into consideration, the fitness of all individuals with A
phenotype is in principle supposed to be the same, and different
from those with phenotype a. Starting from a very simple equation
written by Hardy and Weinberg at the beginning of the twentieth
century, describing the probabilities of the three genotypes in the
absence of all possible factors interfering with reproduction in the
so-called “equilibrium population”, genotype frequencies in subsequent generations can be predicted once allele fitness values have
been calculated. In this operation the environment is implicitly
taken to be constant and to have therefore the same effect on selective levels throughout generations. Although fitness values are
quantitative and the fitness function is implicitly assumed to be
continuous with time, the effect of environment is wholly deterministic as fitness values are constant in time for each allele, the
38
Marcello Buiatti, Marco Buiatti
phenotype being considered simply as a carrier, irrelevant for evolution. Implicit in this extreme view clearly stated in Fisher’s “The
genetic theory of selection”, is the Mendelian principle of lack of
“ambiguity” in the passage from gene to character. Again, as all
calculations are carried out on single genes and the results extrapolated to the whole genotype, this is considered as an assemblage of
independent particles, the Mendelian factors.
However, while selection is thought to be essentially deterministic, neo-Darwinism envisages the presence of two other factors of
allele frequency changes due to pure random events. The first is
mutation, defined as the change of A into a and of a into A, in an
updated version of Hugo de Vries’s concepts. Obviously, mutations will only modify gene frequencies if the probabilities of the
two possible changes are different. The second purely random factor, called random drift, is essentially based on the occurrence of
“sampling errors” in the relative proportions of alleles throughout
generations. Let us take the simplest example of this process,
namely the so-called “founder effect” or “migration”. Suppose we
analyze the behavior of a butterfly population where wing color is
controlled by one gene and two alleles (A = red, a = white). Suppose furthermore that allele frequencies in the home area of this
species where one million butterflies live, is 0.2 for A and, consequently 0.8 for a. Now, let’s imagine that a very small group of 10
butterflies migrates into an island where no individual inter-fertile
with our species exists. As every butterfly has two copies of the
gene for wing color, they have all together 20 alleles. Relative
numbers of A and a, to repeat the situation present in the home
area, should then be 16a vs. 4A. This however happens only in
theory, and only if allele frequencies in the original one million
individuals population have been maintained in the migrant
group. The probability to obtain the same result in the colonized
island, however is very low as, due to the small dimensions of the
migrant sample, small random deviations from the base value will
significantly change future allele frequencies. Therefore, the general neo-Darwinian conception of evolution was based on stochastic
determinism, mutations and random drift being fully stochastic
processes, development during single life cycles being solely determined by heritable information.
Chance vs. Necessity in Living Systems
39
Surprisingly, although all this evidence suggested that living
systems were controlled by key heritable factors, the critical question of their material nature was not asked by geneticists. It was a
physicist instead, Erwin Schrödinger, who, extremely intrigued by
the results obtained by the newly born discipline of Genetics, set
the ground for a new theory of life in a short series of lectures,
assembled in a famous book by the exciting title What is life? published in 1942.
Being a physicist, Schrödinger tried to find analogies between
his discipline and modern Biology. Within this frame, he assimilated the random mutation events to quantum jumps and at the
same time offered a far-sighted explanation of the extraordinary
level of complexity and order of an organism developing from a
single and apparently simple object like a fertilized egg, using for
its “organization” “disordered” matter. It was Schrödinger who
proposed that living systems were able to absorb what he called
“negative entropy”, “extracting” it from the environment through
the exchange of the molecules of food and energy used in metabolism. This way entropy (disorder) would only locally be maintained low, at the expenses of the context where, according to the
second principle of thermodynamics, entropy would be still steadily increasing.
According to Schrödinger the whole process of the reproduction of the ordered organization of life (“order from order” as he
wrote) had to be controlled by an “a-periodic crystal”, the hereditary material, capable of self reproduction and liable to change
with a low frequency due to quantum “jumps”. The control would
be achieved through a “code”, a term with which Schrödinger introduced the concept of “genetic code”, the basis of the “informational theory of life” (Buiatti [1998]) which has been dominating
Biology ever since.
As Schrödinger wrote:
The chromosomes, or probably only a fiber giving the fundamental backbone of what we see under the microscope as a chromosome, are the elements containing in a sort of encrypted code the
entire design of the future development of the individual and of
its functions at maturity………When we compare the chromosome fiber structures to a coded text, we mean that the universal
40
Marcello Buiatti, Marco Buiatti
mind mentioned by Laplace, to which every casual connection is
immediately known, could tell from those structures whether the
egg will develop, in given conditions, into a black cock or a freckled one, a fly, a corn plant, a beetle. …… The expression “code
text” has, alone, a too narrow meaning. Chromosome structures
are, at the same time, the tools used to achieve the development
of which they are symbols. They are coded laws with extensive
power or, to use a different metaphor, they are the project architect and at the same time skilled builders.
For Schrödinger, the unique nature of life, different from other
physical systems, is defined by the “order from order” process,
deterministic and based on the instruction contained in the “aperiodic solid”.
As he says:
The physicist knows that the laws of classical physics have been
modified by quantum theory particularly at low temperatures.
There are many examples of this fact. Life seems to be governed
by a behavior of matter, ordered and controlled by exact laws, not
based only on its tendency to go from order to disorder but, partially, on the conservation of order.
So far Schrödinger, who left to “biochemists” and other biologists the job of the isolation of the “aperiodic crystal” and of understanding the processes leading to the absorption of “negative
entropy” (of order).
We have been discussing and quoting at length Schrödinger
because his thoughts had and still have a relevant influence on
Biology. Three main concepts, proposed by Schrödinger, have in
fact dominated modern Biology for many decades. The first is the
relevance of order in living systems. In Schrödinger’s theory, an
essentially negative meaning had been attributed to randomness as
in the case of mutations, whose frequency, according to Schrödinger himself, has to be maintained low to avoid a too high level of
instability of the systems. This concept of a negative value of variation, present in Mendelian genetics as in the neo-Darwinian
“Modern synthesis”, is fully coherent with the double helix unique
structure of DNA called later the “fundamental invariant” by
Jacques Monod, representing the transmissible key molecule of life
Chance vs. Necessity in Living Systems
41
“necessity”. Also in Monod’s book, probably the most complete
“modern synthesis” of molecular genetics, chance (randomness,
noise) was attributed only to mutations, in general rather rare and
with negative effects, and to the assortment of alleles at reproduction as foreseen by the Mendelians. The two other concepts which
changed Biology have been the notions of “code” and “information”, both related to DNA sequences, whose knowledge should
allow the complete prediction of the phenotype. From this conceptual “corpus” the “informational metaphor” (Buiatti [1998a,
b]; Gagliasso [2000]) and the “reductionist ideology” were later
derived (Buiatti [2004]).
The apex of the success of this very powerful theoretical system
can be considered to have been reached by the proposal of the so
called “Central dogma of molecular genetics” describing the flow
of the information needed for the self construction of organisms,
from DNA to characters. This flow, neatly described in a paper in
Nature by Francis Crick [1958], is considered to be devoid of mistakes and ambiguity and thought to proceed from DNA to a similar molecule, the RNA, through direct transcription, both being
nucleic acids using a four letter (A, T, G, C) code organized into
triplets. The message is then “translated” into proteins, molecules
“written” using a 20 letter code of amino-acids. The translation
uses 61 out of the available 64 triplet combinations of DNA letters, the remaining three being used as “stop” signals for reading.
The fact that the code has been deciphered three years after the
“Central dogma” does not change the value of that proposal but,
rather, helps understanding how it works.
It is worth noting here that not all the concepts contained in
Schrödinger’s theory necessarily lead to a deterministic view of the
“living state of matter”. Particularly, negentropy by itself does not
imply any choice between total order and disorder.
This concept has been utilized by whole sectors of interdisciplinary studies involving at the same time physicists and biologists,
and has opened the way to a series of significant achievements in
the understanding of self-organization processes of “out of equilibrium systems”, both living and not living (see Nicolis and Prigogine [1977], Buiatti [2000], Kauffman [1995], Strogatz [2003]).
This is the area of investigation of contemporary non linear Phys-
42
Marcello Buiatti, Marco Buiatti
ics whose influence on Biology will be the subject of our next
paragraphs. Another difference between Schrödinger’s view and
Modern Biology is his firm opinion that living systems, although
obeying the general laws of Physics, are also regulated by some
specific biological laws like those related with the “order from order” and negative entropy dynamics concepts. This is the reason
why he was also aware that Physics and Chemistry alone would
not be able to completely explain the behavior and dynamics of
life, thus implicitly suggesting the existence of a “living state of
matter” (Buiatti [2001], Buiatti and Buiatti [2004], Buiatti et al.
[2004]).
2. The “Phenotype Paradigm” and the Theoretical Challenge to the
Chance/necessity Antinomy
As mentioned before, classical neo-Darwinists and the molecular biologists of the central dogma attributed only a minor role, in
the history of organisms, to adaptation during their life cycles.
The rationale behind the conceptual removal of any role of the
phenotype in evolution was based on the concepts of gene independence and lack of ambiguity as stated in the central dogma,
both implying the view of the phenotype itself as a single, unambiguous product, of genotype reading without any influence of the
external and internal contexts. It is worth reminding that most of
the often complex mathematical treatments of population dynamics have been based on the analysis and simulations of changes of
frequencies of single gene alleles, consequent rules and “laws” deriving from the extrapolation to whole genomes, envisaged as resulting from the sum of heritable components. Although, however,
the conception represented by the Central Dogma has been largely
prevalent in the twentieth century, the premises for the future,
dramatic changes occurred in the last twenty years in contemporary Biology, were set by a number of scientists already in the days
when the “Modern synthesis” was built, and a new vision of the
living state of matter, although often disregarded, was developed.
The first field in which a role of environmental effects and gene
interactions has been introduced and discussed is a nowadays almost forgotten “sub-discipline” of population genetics, quantita-
Chance vs. Necessity in Living Systems
43
tive genetics, initially based on the discovery of poly-gene systems
controlling quantitative characters by Johannsen and Nilsson-Ehle.
More or less in the same years when the general neo-Darwinist
theory was developed essentially by three geneticists (J.B.S. Haldane, R.A. Fisher, S. Wright), a solution was also formulated for
the other aspect of the Darwinian-Mendelian controversy, the discrete vs. continuous problem. The compromise was reached when
Nilsson-Ehle [1908], working on a quantitative character in barley
(seed color), demonstrated that a continuous distribution of color
intensities in a plant population could be explained with the segregation of a number of Mendelian genes, all giving a positive or
negative quantitative contribution to the chosen character. Color
expression according to this hypothesis was determined by the
numbers of “colored” alleles present in each genotype in a linear
fashion. At the beginning of the new discipline, quantitative genetics, it looked as genetic determinism had won another key battle and that life remained, as in the early years of the century,
mechanistic and controlled by chance and necessity. It is worth
noting here that the success of quantitative genetics was due mainly to the need of new biometrical methods for the prediction of
the effects of intra-specific crossing and artificial selection in domestic animals and plants for productive reasons. Inevitably however, this goal directed the interest of geneticists to the phenotype
and its dynamics during life cycles. The experimental observation
of the dynamics of production relevant characters in subsequent
generations showed that the prediction of phenotype productivity
of individuals on the basis of the performances of the parents was
not reliable as expected due to the effects of the environment. This
observation led to the development of a statistical parameter called
“heritability” used to estimate the relative impact of genetic variability on phenotypic variation. This parameter, generally based
on the use of a hierarchical analysis of variance as well as on parent-offspring regressions and correlations, was found to provide a
good index of predictability of the phenotypic outcome of genotypes. Conceptually, all this means that single genotypes may lead
to more phenotypes each moving during its life along a path, dependent on the environment and its modifications, within a phasespace whose limits are set by the genotype and its plasticity (capac-
44
Marcello Buiatti, Marco Buiatti
ity of phenotypic change).
It was also from the genetics of quantitative characters that
emerged another relevant concept, a new version of the Darwinian
correlated variation.
The Darwinian need, for adaptation and survival, of a certain
level of harmony between the parts of an organism, was proposed
again by quantitative geneticists when they found out that in
many cases selection for one character (for instance meat production in cows) would have negative effects on another (milk production) while in others the interaction was found to be positive.
Fine statistical tools were therefore developed by breeders to measure the relative impact of genotype on correlations allowing modulation of selection pressure to avoid negative side effects and efficiently exploit positive ones.
While the interactions between parts of a living network and
between a living system and the environment were the main discoveries of quantitative genetics, the relevance of non additive interaction between alleles was introduced by the observation in
maize and, later on, in other out-crossing plants and in animals,
that individual “vigor” was highly enhanced in heterozygous organisms derived from crosses between highly homozygous parental
lines. The phenomenon was called “heterosis” and its discovery led
to an abrupt change in plant and animal breeding methodologies
until then aiming at the development through selection of “optimal”, highly homozygous individuals putatively containing in a
“pure” state the “good” alleles. It was the complete failure of a selection program based on self-fertilization in maize that showed
the need to maintain high heterozygosity levels for the improvement of production. Seen with contemporary eyes, this has probably been the first observation of a positive value of variability.
The concept was further supported by the innovative work by
Michael Lerner [1954], one of the most prominent scientists of
neo-Darwinism but also involved in animal breeding, who showed
that hybrid individuals were more stable, during their life cycles,
than the more homozygous counterparts. This result was interpreted by Lerner in his book Genetic Homeostasis [1954] as due to
a sort of “buffering capacity” of variability at the gene level
favoring individuals carrying different alleles in a relevant number
Chance vs. Necessity in Living Systems
45
of genes. In other words, the presence of two different forms of a
gene would increase the homeostatic capacity of individuals
thereby enhancing their phenotypic vigor and fitness. As, obviously, in a given population heterozygosity would be higher in the
presence of a high general level of genetic variation, the natural
extension of the new concept to evolution would lead to the suggestion of a positive value for high variability also at that level.
The concept of “Genetic homeostasis”, the term coined by Lerner
for this process, has later been one of the ground tenets which induced a number of scientists like Richard Lewontin [1972] to introduce the concept of “balanced selection” in Population Genetics emphasizing the positive value of high amounts of genetic variability. This particular development, took place almost twenty
years later of Lerner’s initial work and was also supported by a
large number of experimental data obtained with new biochemical
methods of protein analysis contradicting the classical neo-Darwinist hypothesis. According to that conception, natural selection
should lead to highly homozygous genotypes containing only the
“optimal” allele. Therefore, one would expect to find in present
day species, coming from million years of natural selection, a very
low remaining genetic variability. The opposite was found to be
true with the new methods as most enzymes were found to be
present in the same organism in more than one form. Lewontin
and his followers interpreted this result as deriving from positive
interactions between alleles giving a selective advantage to individuals carrying both of them, homozygotes for each allele showing a lower fitness. The concept of interaction had been extended
to whole genotypes by Sewall Wright, one of the founders of neoDarwinism, who, in a famous debate with R.A. Fisher, proposed
them as the unit of selection, clearly stating that fitness values cannot be calculated on single genes as they derive from the effect of
the non additive combination of many if not all the genes of a
genotype. He then coined the term “genetic landscape” whose
graphic representation was an artificial landscape (a phase space)
with fitness values attributed to each of the many possible gene
combinations. In such a landscape, more than one combination
can be optimal in terms of fitness, and the value of this parameter
changes according to the environmental context.
46
Marcello Buiatti, Marco Buiatti
A completely different explanation of the unpredicted high genetic variability in populations was offered by a group of Japanese
workers led by M. Kimura [1961]. The Japanese interpreted the
high variability observed as being due to the lack of selective forces
acting on a complex set of completely “neutral” variants. The
Japanese “Neutral theory of Evolution” was a new version of the
previously discussed “Mutation theory” of H. de Vries, as Kimura
and his followers, at least in their first period of work, envisaged
again evolution as strictly happening only through mutation and
random drift. The role of the phenotype in this case was not considered, organisms being fully determined by independent genes
once they had been randomly assorted. Needless to say, Kimura’s
theory was in open contradiction with the aforementioned data of
quantitative genetics, not only for what regards the roles of environment, phenotypic variation and selection, but for the refusal of
the concept of interaction itself.
Probably the single person who better described and discussed
the living systems as being in a state somewhat intermediate between order and variation has been Conrad Hal Waddington. He
was an embryologist by formation, with a strong theoretical tendency and interest in mathematics as a source of tools for the description and interpretation of living systems. In the period 19401975, when, due to the successful use of the reductionist method
and to the fast improvement of experimental tools, the pace of
accumulation of analytical data had become very fast, he proposed
what he called (Waddington [1959]) the foundation of “synthetic
Biology” aiming at developing “a coherent intellectual picture of
the whole kingdom of life”.
The most relevant achievements of Waddington’s synthetic
Biology program are the results of a series of interdisciplinary
meetings which produced four volumes by the title “Towards a
theoretical biology” from 1969 to 1972. Many of the concepts and
ideas in Biology proposed at that time were anticipatory of the
discoveries of the last fifteen years or so and resulted from intense
and fruitful collaboration and discussions between geneticists,
physicists, embryologists, and mathematicians. It was through
these discussions that Waddington developed what he called the
“Phenotype paradigma” aimed at complementing the neo-Darwin-
Chance vs. Necessity in Living Systems
47
ist view and at a better characterization of the specific character of
the “living matter”.
He started considering that “one of the most basic characters of
living systems, considered as active machines, is the fact that they
produce a local enhancement of order through the assumption of
simple molecules and their organization into complex compounds,
orderly structured. At first sight one could think that those systems act in opposition to the second law”. Although surprisingly
Waddington very seldom quoted Schrödinger, this statement is
fully coherent with the “negentropy” theory proposed by the
physicist. However, Schrödinger’s concepts are further implemented by Waddington with a thorough discussion on the real
value of the application of information’s theory to life.
His attitude was highly critical towards the direct use of information theory particularly in the case of information transfer from
DNA to protein and to characters during development:
In this case we find that Biology has developed more flexible
mechanisms than those used by telephone technicians. There is a
system through which information can be changed in the sense
that a gene can mutate so that what the progeny receives is not
exactly what was present in the parents… It’s in the later phases
of phenotype formation that information theory fails to deal with
the situation. It is obvious that one organism’s phenotype is not
simply a collection of all the proteins corresponding to all the
genes of the genotype. It is true that the phenotype is the result of
the assemblage of parts highly heterogeneous, in each of which
some but not all the proteins are present which would derive from
the genes, but also many structures and substrates other than primary proteins corresponding to specific genes …. It appears completely obvious following the “common sense” that a rabbit running in a field contains an amount of “variation” higher than that
contained in a rabbit’s egg immediately after fertilization. Now,
how can we explain such a situation within the frame of information theory whose fundamental principle states that information
cannot be increased… in fact the main problem is that throughout the passage from a zygote to an adult organism, “information”
is not simply transcribed and translated but works in terms of “instruction” or, to say it in a more elegant manner, of “algorithms”.
DNA produces RNA which in turn builds a protein.
48
Marcello Buiatti, Marco Buiatti
The protein then affects the surrounding components in different
ways, whose result is the production of a higher variety of molecules than that present at the beginning”.
Of course, this is possible, as observed by many authors, and also
by Waddington himself, because living systems are open in the
sense that they continuously exchange energy and matter with the
context around them. The number, quality and configurations of
the components of biological systems are continuously modified
following complex non additive interaction processes within the
system itself and with the ever changing environment. Complex,
non linear functions are needed to describe organismal paths of
self-organization based on the Waddingtonian principle of homeorrhesis, that is the dynamic counterpart of homeostasis or, from
the Greek, a flow in which the system maintains its “self”. The
paths followed by single individuals have been called by the British
scientist “chreods”, each of them being one of the potential life
cycles of the organisms allowed by the “inherited tools” and by the
contexts. While homeostasis can be defined as the capacity of a
system to maintain constant its main features, responding through
a series of buffering processes to the noise coming from the external context, homeorrhesis is its dynamical version, that is the
property of living beings to avoid diverting their paths from the
system-specific attractor through a continuous exercise of adaptation through change. While an organism absorbs “order” from the
context, as foreseen by Schrödinger, it also increases disorder/information all through its life cycle.
Seen this way, developmental but also evolutionary paths do
not seem directed in any way towards the stable acquisition of an
optimal order as it happens with crystals and human-made machines. Rather, they appear to obey to homeorrhetic laws maintaining specific equilibria between order and change within the
frame of the requirements of the second principle of thermodynamics.
Implicit in the non additive nature of the interactions among
components of living systems and between organisms and the environment, is the presence of non linear components in functions
describing them. It is worth noting that a relevant role in this re-
Chance vs. Necessity in Living Systems
49
spect, has been played by the friendship and cooperation between
Waddington and René Thom. The French mathematician described from a topological point of view the dynamics of developmental evolutionary processes on the basis of Waddington’s biological suggestions, thus rendering the whole theory clearer and inducing a new interest in the use of mathematics in biological studies beyond the probability theory and statistics dear to many neoDarwinists. Particularly, the reciprocal exchange of views and data
between the developmental geneticist and the mathematician had
probably an influence on the development by Thom of the socalled “Catastrophe Theory” and the introduction of the notion of
bifurcation, describing sudden changes in systems reaching a critical point, directly reminiscent of chaos theory. A well known heuristic representation of these concepts by Waddington and Thom
can be seen in the “epigenetic landscape”, i.e. the attractor within
which a living system moves in development and evolution. Such
a landscape shows the minima in which a path will be running,
allowing the “jump” from one valley to another when the dynamic
system reaches a bifurcation point.
It is worth noting that, while Waddington and the relatively
small group of theoretical biologists who followed him seemed to
be rather isolated from the mainstream of experimental life sciences, their conceptual framework was enriched, in the seventies
and in the eighties of the twentieth century, by the work of physicists and mathematicians like those of the groups led by Prigogine
(see Prigogine and Nicolis [1972]), Henry Atlan (Atlan [1979]),
Maturana and Varela [1980], Goodwin, and many others. A particularly interesting result of their work was the extension of
Schrödinger’s intuitions and the demonstration and application to
life of the concept of self-organization. Prigogine’s work was based
on the observation coming from chemistry, that systems out of
equilibrium, or, in other words, embedded in a matter/energy
flow, can undergo a transition to order which is maintained until
the recovery of an equilibrium state or a chaotic transition. A first
mathematical modeling of such systems goes back to Alan Turing
[1952] but has been carried out much further by Prigogine and his
co-workers who introduced and modeled the concept of dissipative
systems and structures. The concept of self-organization, implicit
50
Marcello Buiatti, Marco Buiatti
in Waddington’s work, had been introduced quite early by J.
Needham in his famous book “Order and life” and was later taken
up by Waddington himself and the students of biological complex
systems. It was then suggested that dynamical interactions among
components may be the main source of constraints, keeping the
variation of systems during development and evolution within the
attractors as defined earlier on in this paper. A particular attention
was later on given to this area of thought by the whole group of
Santa Fe starting from S. Kauffman’s work (Kauffman [1995]).
Kauffman’s approach is essentially based on simulations of the
dynamics of N/K Boolean systems, containing N components with
K connections. The simulations have be implemented by introducing connection functions and the result of computer simulations
can be treated with statistical methods describing the dynamics of
fluctuations. Kauffman’s simulations generally did not involve experimental validation. Recently, a different approach to the study
of what has been called “spontaneous order” involving “Sync”
(synchronous behavior) has been used (see Strogatz [2003]). This
new way of looking at self-organization of living matter was introduced by a pioneering work by Watts and Strogatz (Watts and
Strogatz [1998], [1999]) and is based on the study of the networks
formed by the interactions among the components of real living
systems. According to Strogatz, their research had been prompted
by the observation of the mixture of order and randomness of web
sites network and of the strange way of spreading of infection diseases.
Strogatz and Watts and almost at the same time the group of
Barabasi (Barabasi [2000]) started looking at a large amount of
heterogeneous networks studying the distribution of numbers of
connections of the different components. The networks analyzed
included a variety of examples such as the systems of actors present
contemporarily in movies, of genes in the genotypes of a worm
and a yeast, nervous systems etc. and they surprisingly showed always the same basic structure. All these networks comprise a very
high number of components each connected with others with a
variable number of links. When one studies the shape of the curve
describing the statistical distribution of the number of links per
node, he will always find that it is not Poisson, as it would be ex-
Chance vs. Necessity in Living Systems
51
pected for random interactions, but a power law, denoting a scalefree structure of the network. This implies that these networks
have a very inhomogeneous distribution of links, allowing the simultaneous presence of a few nodes linked to many other nodes
and a large number of poorly connected elements (Boccaletti et al.
[2006]). Such scale-free structure allows a fast information flow
even between distantly related components, because the number of
links between any two components is always very low.
A parallel current of thought (H. Atlan, J. Wicken, J. Brooks
and J. Wiley etc.) looked at the problem from another point of
view, namely the effect of variation and constraints on entropy
functions. Thus, for instance, H. Atlan, proposed to consider the
calculated “observed” Shannon entropy of experimentally analysed
biological systems as the result of constraints determined by selforganisation processes reducing the “potential” entropy expected
in the case of a totally random distribution of the n components)
These concepts contradicted the false alternative between chance
and necessity in an explicit way and opened the way to a very
fruitful area of studies aiming at better statistically defining the
“living state of matter” (Buiatti and Buiatti [2004]) and understanding the functional dynamics of constraints.
All the just discussed interdisciplinary work along with earlier
theories such as those by Lewontin, Lerner, Waddington, Wright
and many others including those implicitly derived from quantitative genetics, have been often emarginated from the mainstream of
discussion in experimental Biology until the end of the last decade
of the twentieth century and the beginning of the third millennium. Surprisingly, the debate on self-organization, and in general
on the just discussed concepts was, on the other hand, very wide
in non biological areas such as those of mathematics, physics, philosophy and involved also a part of the large public, interested in
the problem of the level of predictability of the dynamics of complex systems. Unfortunately, a large majority of hard core biologists for a long time either did not read or psychologically removed the major inputs by physicists and mathematicians working
in the area of non linear dynamics. The experimental area was in
those times completely dominated by what we would call the
“Central dogma ideology”, whose biased interpretation induced
52
Marcello Buiatti, Marco Buiatti
life science students to privilege studies of the “ordered side” of life
and led to the interpretation of variations as disrupting white-noise.
As it happens in science, however, the experimentalists had
soon to cope with the extraordinary amount of data coming from
the use of the powerful reductionist method often showing an increasing amount of “exceptions” to the common paradigm.
In the next paragraph therefore we will try to summarize some
relevant achievements of the new, contemporary, Biology, certainly not with the aim to build a new paradigm but to update the
state of the art of the present vision of the “living state of matter”
(Buiatti and Buiatti [2004]). Particularly we shall report on the
ever increasing evidence of “friendly” co-existence in living beings
of a number of “variability generators”, fixed by evolution, and the
“spontaneous order” derived from interactions between components. We will then discuss the thesis that the “disorder” generated
but contained is “benevolent” in the sense that it allows the maintenance of the connections necessary for life, contrasting the effects of the disrupting internal and external random noise.
3. THE CONTEMPORARY REVOLUTION IN BIOLOGY
AND “THE BENEVOLENT DISORDER”
Before entering into the discussion of recent experimental data
leading in our opinion to the solution of the chance vs. necessity
antinomy, we shall introduce a few concepts that we consider relevant to the comprehension of the general framework of the debate on the “living state of matter” of the last decade of the XXth
century and the beginning of the third millennium.
In the first place, recent data show that different groups of organisms, namely prokaryotes, plants and animals and humans display different adaptation strategies although they are all “explorative”.
The term “Explorative strategies” has been proposed in 1997 by
Gerhart and Kirchner meaning by that adaptive strategies based on
the exploration of variability generated by specific “generators”
(Buiatti et al. [2004], Buiatti and Buiatti [2004]) and on selective
processes, developed during evolution, allowing the “choice” of the
Chance vs. Necessity in Living Systems
53
specific variants needed to answer to incoming, disruptive noise
and in general to changes in the context. In all cases therefore, a
certain level of “disorder” (variability) is needed but shall always
be limited by constraints due to the need of a certain level of “internal harmony” in living dynamical networks. The variability induced by generators can be heritable or limited to the phenotype.
Heritable variation, as proposed and discussed at length in Jablonka and Lamb [2004] may be genetic, epigenetic, behavioural, symbolic. In all cases heritable variation of populations, species, and
ecosystems will contribute to a measurable parameter also proposed
by Gerhard and Kirschner called “Evolvability” or, following Earl
and Deem, “propensity of evolve” (Earl and Deem [2004]), that is
the capacity to adapt throughout generation through structurefunction changes. Evolvability levels are determined by inborn
processes leading to genetic variation fixed through evolution for
their value “per se” as shown by the fact that variations increasing
this parameter bear in themselves a selective advantage.
However, epigenetic variation, is, for animals and plants the
major source of phenotypic adaptation and can be in some cases
transmitted to the progeny without changes in DNA sequences.
The term “epigenesis” has been introduced, as mentioned earlier, by C.H. Waddington, and epigenetic plasticity in a very broad
definition, includes all processes allowing variation during life cycles not already present at birth as information contained in the
DNA. As shown by several authors also plasticity can be selected
as such thus increasing adaptation capacity based on the development of a series of tools for the exploitation of “ambiguity” (more
than one meaning of information stored in nucleic acids and proteins, utilized according to different codes).
Finally, as discussed before, it is now universally accepted that
living systems are dynamic networks endowed with different levels
of robustness (the capacity of a system to resist to incoming random noise through the use of its internal plasticity), redundancy
(the presence of more than one copy of living system components)
and vicariance (the presence of alternative pathways allowing the
achievement of the same result). The very notion of network immediately recalls the Darwinian intuition partially confirmed by
quantitative genetics, of the presence of correlations between the
54
Marcello Buiatti, Marco Buiatti
parts of an organism (the correlated variation). The meaning of
the Darwinian principle was that the parts are not independent
one from another in the sense that none of them is free to assume
all the “configurations” possible (sizes, weight, shapes and so on)
but that only those changes are allowed which do not harm the
overall “harmony” of the system. To translate this concept into
modern terms, we can simply extend it to all components of living
systems at all levels of the hierarchical organization of life. Communication within networks is in fact the reason why not all the
variability induced by the aforementioned “generators” can be utilized by living systems, the “benevolent” part of it being composed
of elements harmonically inserted in the communication network.
One may recall here the concept of “internal selection” proposed
by Bateson (Bateson [1980]). By internal selection he meant that,
as the components of living systems are part of a connected network with its harmonic quantitative and qualitative rules, even a
single mutation may damage not only the affected component but
a number of others connected with it thus eventually decreasing
the fitness of the organism.
Having thus defined some basic interdisciplinary concepts of
contemporary Biology, let us now start discussing, using the existing evidence, our proposal of solution of the chance vs. necessity
antinomy based on the contemporary presence of variability on
one hand, and the choice of different portions of it on the other,
through communication between components of living systems
and with the environment. In other words, in our opinion, life is
maintained by ever changing equilibria between the “benevolent
order” and the needed amount and quality of the “benevolent disorder”. We will now provide evidence on how benevolent order
and benevolent disorder emerge in DNA, the hereditary material
and therefore the “living memory” of biological dynamics, liable to
be statistically characterized as a string of information. This is now
possible as whole genome sequencing has being carried out in hundreds of eukaryote and prokaryote species and has completely
changed our conception of genome structures and functions. Genome size has been shown to vary from a few thousand to billions
of nucleotides and to be much smaller in prokaryotes (in the order
of millions) than in eukaryotes (in the order of billions). The in-
Chance vs. Necessity in Living Systems
55
crease in DNA amounts in plants and animals moreover has been
shown to be due to a far higher amount of non-coding sequences
than coding ones. We know now that coding sequences cover
80% or more of prokaryote genomes but less than 2% for instance
in the humans. A bacterium like E. coli has about 4000 genes, in a
genome of four millions while the human genomes only have
23000 according to the last estimates although their total DNA
content amounts to more than 3 billions nucleotides. So, there is
in this case an inverse relationship between the complexity of the
organisms and the ratio coding/non coding DNA.
Before providing evidence on generators of “benevolent disorder”, we first focus on the widespread ‘benevolent order’ in noncoding DNA sequences, since as we will explain later on, the two
benevolent components are probably tightly linked. “Order” in
non-coding DNA sequences emerges from several studies on the
extent of constraints to randomness in DNA strings. As it often
happened in the history of life sciences, physicists and mathematicians involved in the general discussion on this problem in nature
were the first to enter in this research field using a range of analytical methods such as the study of periodicities, short and long
range correlations, or information theory measures (Li [1997]).
Interestingly, the picture coming from the results obtained with
the different methods used in the analysis of the fast increasing
DNA sequences available from gene Banks is quite coherent. In
general, constraints to randomness are higher in eukaryotes than
in prokaryotes and, particularly in the first class of organisms, in
non coding sequences than in coding ones. While until the end of
last century the prevalent opinion was that non coding DNA was
thought to be devoid of any significant functional meaning and for
this reason was called “junk DNA”, the analyses of whole genomes
transcribed RNA products (the “transcriptome”) carried out in the
first years of the third millennium have shown that a large part of
non coding DNA including repeated sequences is indeed transcribed into RNA and therefore has to be considered functionally
active. The fact that the distribution of nucleotides is far from randomness in eukaryotic non coding DNA can be attributed to several experimentally observed causes. As discussed more at length in
Buiatti et al. [2004], eukaryote genomes contain large numbers of
56
Marcello Buiatti, Marco Buiatti
short and long sequences present in many copies (repeated sequences) whose function is often as yet largely unknown. These
are generally classified into long and short repeats according to
their length. A large part of the long repeated sequences in eukaryotes
are “transposons”, the mobile elements discovered in 1951 by
Barbara Mc Clintock in maize, and afterwards forgotten for a very
long time, mainly due to the contradiction between their existence
and the dogma of DNA stability We know now that they often
cover more than 50% of eukaryotic genomes and are responsible,
for instance in maize, for “catastrophic events” of genome shuffling recently discovered (Brunner et al. [2005]). Short repeated
sequences are composed of a variable number of identical nucleotides,
of purine or pyrimidine, “strong” (G, C) or “weak” (A, T)
nucleotides, or of repeats of 1-7 nucleotides called “microsatellites”.
All these sequences have been called “homogeneous” or “low complexity” and their distribution in the genomes has been found not
to be random (Buiatti et al. [2004]). Moreover, again in eukaryotes,
genes themselves are not randomly distributed along the chromosomes but concentrated in specific regions, and the GC/AT ratio
is heterogeneous as shown long time ago by Bernardi and his
group (Bernardi et al. [1985]), who called the homogeneous areas
“isochores”. Finally, regions upstream and downstream of eukaryotic
exons (the eukaryote coding sequences), contain specific “motifs”
highly conserved throughout evolution, often at regular distances
from the beginning or the end of transcription units.
The main question to be asked at this point is whether or not
constraints to randomness found in the non-coding eukaryotic
sequences are “signaling” specific functions, and therefore their
nucleotide composition and order acts as a sort of “hidden code”,
as suggested by Trifonov [1998]. We will now describe experimental data suggesting that this is probably often the case, and that
many of these functions are generators of “benevolent disorder” or
tools for choosing the portion of variation to be utilized for adaptation in different contexts.
Before discussing these data, however, it should be noted that
the different amounts and distributions of constraints to randomness in prokaryotes and eukaryotes may be related to the different
adaptation strategies adopted by the two groups of organisms. We
Chance vs. Necessity in Living Systems
57
now know that while in prokaryotes adaptation mainly relies on
genetic variability, processes leading to phenotypic plasticity are by
far more frequent in plants and animals. The different strategy
seems to derive from the very structures/functions of the prokaryotes and eukaryotes. Particularly, bacteria have, as individuals,
very short life cycles and are haploid, that is endowed with only
one copy of each gene. Now, short life cycle means a very high
capacity to spread in a bacterial population a mutation conferring
selective advantage. This process is also easier because of the haploid nature of the bacterial genome, allowing the immediate expression of the new allele, independently of its recessive or dominant nature. On the other hand, eukaryotes are endowed with
genomes at higher ploidy levels, have by far longer lives, and therefore
need several long-lasting generations for a favorable, new mutation
to be expressed and fixed in a population.
Consequently, bacteria have developed during evolution a number
of very sophisticated processes for the exchanges of significant
portions of genomes both at the intra-specific and at the inter-specific levels. In fact, exchanges are so frequent that the very concept
of species is rightly being questioned in the case of microorganisms. Moreover, bacteria are also endowed, as shown recently
(Lombardo et al. [2004]), with systems for stress perception and
consequent activation of “mutator” genes increasing mutation
frequencies and therefore the probability of occurrence of mutations favorable for response to stress. This is the case of the socalled RpOs system: upon stress, the RpOs gene is activated, inducing the expression of mutator genes like those coding for DNA
polymerase IV, an inducer of DNA replication with mistakes. In
other words, in bacteria, a process has been developed during evolution through selection leading not to the fixation of specific
mutations, but increasing the probability of obtaining them
through higher mutation frequencies (new “benevolent disorder”
available).
It should be noted that in this case, the process of variability
generation is initiated by recognition of stress signals by the cell
and eventually by the non-coding DNA sequences upstream of the
RpOs gene. RpOs then produces a protein which behaves like a
signal to the mutator genes again through recognition. Recogni-
58
Marcello Buiatti, Marco Buiatti
tion is necessary for the activation of all DNA functions, both in
prokaryotes and eukaryotes. For instance, both replication and
transcription are induced and regulated by signals, whose recognition generally occurs through the formation of complexes with
DNA in well localized positions of the string. Complex formation
occurs whenever local DNA conformations are complementary
with signaling proteins “shapes” and charge distribution,
complementarity being the limiting factor of recognition at all levels in living systems from the enzyme-substrate complexes, to
RNA-protein, RNA-RNA, RNA-DNA, protein-DNA, proteinprotein complex formation. All this happens because all macromolecules, DNA included, are endowed with “conformation landscapes”, that is the “ensemble” of possible conformations that a
molecule can assume, including those placed at “energy minima”.
So, DNA, far from being always a neat double Watson Crick helix, can have many shapes each of which being determined by the
specific DNA sequence composition, by the physico-chemical context, by possible pre-existing complexes with other molecules of all
sorts. And, indeed, regulation occurs because many local conformations and the relative local sequences have been selected for, to
allow localized and function-specific complex formation. In other
words, at least a part of DNA sequences constraints (deviations from
randomness) are determined by the functional role of complex
formation with other molecules equally constrained. In the case of
RpOs, constraints in DNA sequences leading to recognition are
needed for the release of adaptive “benevolent disorder”.
In eukaryotes, although genetic variability is certainly a tool for
long term evolution at the population and species level, adaptation
during life cycles relies on phenotypic plasticity, a part of which,
however is due to somatic genetic changes. It has been shown, in
fact, that mutation frequencies are not random in eukaryote genomes,
being higher in the already mentioned “low complexity sequences”,
and that these sequences are often localized in parts of the genomes
where high mutation rates are requested. This is the case, for instance of the sequences coding for antigen/antibody recognition in
immuno-globulin genes, which mutate by shuffling of repeated
tracts during cell development. This process leads to the production of an extremely high number of different antibodies in single
Chance vs. Necessity in Living Systems
59
individuals conferring the capacity to recognize an extraordinary
number of “enemies” (antigens) (see for instance Lewin [2007]).
Moreover, it is known that transposons also in eukaryotes may be
activated in the presence of stress (Madlung and Comai [2004])
and will thus increase somatic mutation frequencies, by “jumping”
from one part to another of the genome, thus determining shuffling of fairly long sequences. All these processes, generally occurring at the somatic level in eukaryotes, are examples of the apparent paradox of constraints to randomness in DNA sequences leading to increases in variability.
However, the main processes leading to phenotypic plasticity in
eukaryotes are the differential activation and modulation of gene
sets in different internal and external contexts and gene “ambiguity” obtained through signal recognition.
To give an idea of the levels of plasticity that can be reached by
gene differential activation and modulation, humans are endowed
with about 23.000 genes, around 2000 of which are thought to be
active in every single cell. However different cells, differently located in the human body “use” different gene combinations and
the same is true for single cells in different periods of their life
cycles. Moreover, gene activity far from being a binary, yes or no,
process is modulated through changes in the stability of the socalled “transcription complex” between DNA and a large number
of proteins. To obtain a stable transcription complex DNA has to
curve, twist and roll at specific points, these conformation changes
depending on DNA flexibility, only possible in the presence of
specific sequences in the right positions along the string. Activation patterns can be transient or stable for even long periods of
time. The stabilization mechanism generally involves methylation
or de-acetylation processes induced by one or more protein-DNA
recognition events through shape complementarity. Gene ambiguity, that is synthesis of more than one protein from single genes,
on the other hand derives from the use of different starting and
ending sites in the DNA “reading” process as in the case of the socalled “overlapping genes” discovered in prokaryotes but much
more frequent in eukaryotes, or through “alternative-splicing” of
transcribed RNA. The last process occurs only in eukaryotes where
genes are discontinuous in the sense that they are endowed both
60
Marcello Buiatti, Marco Buiatti
with coding (exons) and non coding (introns) sequences.
Gene expression starts with the transcription into RNA of the
whole sequence but later transcribed introns are dismissed and destroyed, exons being assembled to yield the “mature DNA” on the
basis of which proteins will be synthesized. However, exons can be
assembled in different orders leading to a number of different sequences and, consequently, to the synthesis of more than one protein. Recent data suggest that an average of 10-15 proteins per
gene are synthesized on the basis of the 23,000 human genes but
the ambiguity level varies according to protein function. For instance neurexins, proteins contributing to the connection between
human neurons, are more then 2000, but they are coded by only
three genes; similarly, in Drosophyla only one gene has been shown
to produce 38000 different proteins. A higher ambiguity capacity
in proteins of the human brain is coherent with the general human adaptation strategy mainly based on “inventions”. In all these
processes, the “choice” of the proteins to be produced in every
spatially and temporally variable context relies on signal recognition and transduction through intermolecular complex formation.
Processes leading to ambiguity are not limited to transcription
and protein synthesis. Functional proteins are often the result of
the assemblage of a number of polypeptide chains allowing more
than one combination, and therefore different “meanings” (functions). Moreover, each protein has its own conformational landscape, resulting in different minima and functions. One typical
example of this is the “mad cow” illness (BSE), where two different conformations of exactly the same protein make the difference
between life and death for the cow itself and the humans eating it.
Calmodulin is another often quoted example as it can form at least
30 different complexes with other proteins leading to different
conformations and functions. Finally, an already assembled protein may complex with other compounds such as for instance sugars and lipids and again hence perform different functions. Again,
inter-molecular communication and complex formation increase
the “benevolent disorder” at the somatic level and at the same time
limit the range of combinations possible between different molecules,
allowing the choice of the proper ones for each specific temporal
or spatial context. Communication therefore is a key process for
Chance vs. Necessity in Living Systems
61
all living systems at all levels of complexity, from the molecular
one to cells, organisms, populations, species, ecosystems and the
biosphere, the robustness of each level of organization being correlated with the variability potential of the lower one and at the
same time with the efficiency of communication between components of the modular networks previously discussed allowing the
exploitation of vicariance and redundancy capacity of the systems.
How close is the link between variability generators and the
choice of the necessary part of variation for adaptation on one
hand and constraints to randomness in the DNA code on the other?
Biologically functional constraints leading to reduction in entropy measures and, particularly, to short and long range correlations in DNA sequences are manifold and therefore we shall just
mention a few examples.
Causes of constraint fixation can be structural and functional.
To the first class belong periodicities deriving from DNA strings
compression through DNA/protein complex formation into
nucleosome and different levels of winding of the double helix as
discussed for instance by Arneodo et al. [2001]. At the functional
level it is known that four dimensional local DNA conformational
landscapes are critical for signal recognition and modulation of
gene expression. Particularly, specific “motifs” are located in all
eukaryotic genes upstream, in the non coding internal part (introns)
and downstream; in mammals GC rich regions are near to transcription initiation sites and in all cases strings located at different
distances from the mRNA production starting point, must be endowed with specific patterns of twisting, bending and rolling to
obtain optimal levels of expression according to the contexts, in
coherence with the work by Audit et al. [2001] who showed a link
between the distribution of DNA bending profiles and long-range
correlations in eukaryotic DNA sequences. Finally, levels of sequence stability and conservation play a relevant role in modulating both plasticity and homeostasis in different regions of the genome leading to the formation of AT and GC patches on the
strings. It is plausible to imagine therefore that such constraints
may extend for long tracts in the sequence, and give a statistical
signature in terms of correlations of that sequence. However, the
number of processes leading to constraints to randomness is enor-
62
Marcello Buiatti, Marco Buiatti
mous and many of them are still to be discovered. Future studies
building a large-scale framework that includes a large number of
these processes into a biologically plausible physico/chemical
model of DNA may improve our insight on the complexity of the
DNA system.
4. CONCLUSIONS
The data just reported seem to us to solve the antinomy between chance and necessity showing that living systems are always
at an intermediate state between total randomness of the distribution of the components and total order. All living beings need
therefore the “benevolent disorder” (genetic variability and phenotypic plasticity) to actively react to internal and external noise
during single life cycles and throughout evolution. Both evolvability
and plasticity are critical for survival, the first being more relevant
for prokaryote adaptation, while the second being the key for
eukaryote lives. During evolution, refined and different tools have
been developed in the two domains through selection of both
evolvability and plasticity.
Genes and alleles increasing phenotypic plasticity and evolvability
then, at the population level, are a part of the gene pool or, to say
it with the words of Sewall Wright, of the genetic landscape (the
genetic space) in which populations and species are situated. The
larger the genetic space of a population the higher its evolvability,
the larger the phenotypic space of individuals in that population
the higher their individual fitness, leading as mentioned before, to
the increase in plasticity related coding and non coding sequences
during evolution.
The second point worth being mentioned here is the fact that
the network structure of living systems and the consequent “internal selection” pressure may involve acceleration of the evolutionary
dynamics when a non-lethal but major change occurs in a network. This is the case of the “hopeful monsters” proposed by R.
Goldschmidt, that is organisms where such a mutation has occurred and selection will be acting (this at least is the “hope” of
the “monsters”) to adapt the rest of the genome to the effect of the
Chance vs. Necessity in Living Systems
63
mutation by selecting a new, more harmonic combination of genes
and alleles.
It should be stressed here that the proposed new vision of evolutionary processes does not by any means disprove neo-Darwinism and its all currents of thought because certainly it is true that
allele frequencies vary due to selection, mutation and random
drift. However, the new data show that the Modern Synthesis does
not take into account several key factors and processes not sufficiently considered so far, suggesting the need to return to the ideas
of Charles Darwin himself although with a much deeper knowledge of the material basis of the different processes.
Finally, we think that what we have discussed is only part of
the present day revolution and that in the future it will be necessary to take into account the “four dimensions” of evolution introduced by Jablonka and Lamb [2004], namely the genetic, epigenetic, behavioral, symbolic.
1
2
Dept. of Evolutionary Biology, Firenze University, Firenze (Italy)
Center for Mind Brain Sciences, Trento University, Trento (Italy)
REFERENCES
Atlan, H. [1979], Le crystal et la fumée, Seuil, Paris.
Audit, B., C. Thermes, C. Vaillant, Y. d’Aubenton-Carafa, J.-F. Muzy and A.
Arneodo [2001], Long-range Correlations in Genomic DNA: A Signature of the
Nucleosomal structure. Phys. Rev. Lett. 86: 2471-2474.
Barabasi, A.L. [2002], Linked, the New Science of Networks. MIT Press, Cambridge
(Mass.).
Bateson, G. [1980], Mind and Nature. McMillan Publ., New York.
Bernardi, G., B. Olofsson and J. Filipsky [1985], The Mosaic Genome of Warm
Blooded Vertebrates. Science 228: 953-958.
Boccaletti, S., V. Latora, Y. Moreno, M. Chavez, D.-U. Hwang [2006], Complex
Networks: Structure and Dynamics. Physics Reports 424: 175-308.
Brooks, D.R. and E.O. Wiley [1988], Evolution as Entropy. Chicago U.P., Chicago.
Brunner, S., K. Fengler, M. Morgante, S. Tingey, A. Rafaski [2005], Evolution of
DNA Sequence Nonhomologies among Maize Imbreds. The Plant Cell 17: 343360.
Buiatti, M. [1998], L’analogia informatica del dogma centale e le conoscenze attuali
della biologia. In L’informazione e le scienze della vita. Angeli, Milano.
Buiatti, Marcello, Marco Buiatti [2001], The Living State of Matter. Riv. Biol./Biol.
Forum 94: 59-83.
64
Marcello Buiatti, Marco Buiatti
Buiatti, M. [2004], Il benevolo disordine della vita. UTET, Torino.
Buiatti, M., P. Bogani, C. Acquisti, G. Mersi, L. Fronzoni [2004], The Living State
of Matter: between Noise and Homeorrhetic Constraints. In Nonextensive Entropy - Interdisciplinary Applications. M. Gell-Mann and C. Tsallis (eds.), Oxford U.P., Oxford.
Buiatti, Marco, Marcello Buiatti [2004], Towards a Statistical Characterisation of
the Living State of Matter. Chaos, solitons and fractals 20: 55-61.
Crick, F. [1958], Central Dogma of Molecular Biology. Nature 227: 561-563.
De Vries, H. [1901-1903], Mutations Theorie. Veit, Leipzig.
Earl, D.J. and M.W. Deem [2004], Evolution is a Selectable Trait. Proc. Natl.
Acad. Sci. 101: 11531-11536.
Gagliasso, E. [2001], Verso un’epistemologia del mondo vivente. Guerini, Roma.
Gerhart, M. and J. Kirschner [1997], Cells, Embryos and Evolution. Blackwell Science, Malden (Mass.).
Goldschmidt, R.A. [1940], The Material Basis of Evolution. Yale U.P., Yale.
Goodwin, B. [1995], How the Leopard Changed Its Spots. Weidenfeld and Nicholson, London.
Huxley, J. [1942], Evolution, the Modern Synthesis. Harper, New York.
Jablonka, E. and M. Lamb [2004], Evolution in Four Dimensions. MIT Press, Cambridge (Mass.).
Jacob, F. [1970], La logique du vivant. Gallimard, Paris.
Kauffman, S. [1995], At Home in the Universe. Penguin Books, London.
Kimura, M. [1961], A Mathematical Problem in Population Genetics. Proc. Camb.
Phil. Soc. 57: 574-582.
Lerner, I.M. [1954], Genetic Homeostasis. Oliver and Boyd, London.
Lewin, B. [2007], Gene IX. Prentice Hall, New York.
Lewontin, R.D. [1974], The Genetic Basis of Evolutionary Change. Columbia U.P.,
New York.
Li, W. [1997], The Study of Correlation Structures of DNA Sequences - A Critical
Review. Computers & Chemistry 21(4): 257-272.
Lombardo, M.J., I. Aponyi, S.M. Rosenberg [2004], General Stress Response Regulator RpoS in Adaptive Mutation and Amplification in Escherichia coli. Genetics
166: 669-680.
Madlung, A. and L. Comai [2004], The Effect on Stress on Genome Regulation
and Structure. Annals of Botany 94: 481-495.
Maturana, H.R. and F.J. Varela [1980], Autopoiesis and Cognition. Reidel Publishing Company, Dordrecht.
Mayr, E. [1982], The Growth of Biological Thought, Diversity, Evolution and Inheritance. Harvard U.P., Cambridge (Mass.).
Mc Clintock, B. [1951], Chromosome Organisation and Gene Expression, Cold
Spring Harbor. Symp. Quant. Bio. 16: 13-47.
Monod, J. [1970], Le hazard et la nécessité. Seuil, Paris.
Needham, J. [1986] (1941), Evolution and Thermodynamics. In Time, the Refreshing River. Russell Press, Nottingham, pp. 207-233.
Nicolis, G.G. and I. Prigogine [1977], Self-organisation in Non Equilibrium Systems. John Wiley, New York.
Nilsson-Ehle, J. [1908], Einige Ergebnisse von Kreuzungen bei Hafer und Weizen.
Chance vs. Necessity in Living Systems
65
Bot. Notiser 1: 257-294.
Orel, V. [1983], Mendel’s Achievements in the Context of the Cultural Peculiarities of Moravia. In Gregor Mendel and the Foundations of Genetics. Mendelianum, Brno, pp. 23-47.
Schrödinger, E. [1942], What is Life? The Physical Aspects of the Living Cell. Cambridge U.P., Cambridge (Mass.).
Strogatz, S.H. [2001], Exploring Complex Networks. Nature 410: 268-272.
Thom, R. [1979], Stabilité structurelle et morphogenèse. Inter-Editions, Paris.
Trifonov, E.N. [1989], The Multiple Codes of Nucleotide Sequences. Bull. Math.
Biol. 51: 417-432.
Waddington, C.A. [1974], A Catastrophe Theory of Evolution. Ann. N.Y. Acad.
Sci. 231: 32-42.
Wicken, J. [1987], Evolution, Thermodynamics and Information. Oxford U.P., Oxford.
Wright, S. [1988], Surfaces of Selective Value Revisited. Am. Naturalist 131: 115123.